A New DTI Image Denoising Method Based on Shearlet Shrinkage and Complex Diffusion

The diffusion tensor image (DTI) is clinically polluted by Rician noise, which can bring serious impacts on tensor calculating, fiber tracking and other post-processing. To decrease the effects of the Rician noise, this paper presents a new DTI denoising scheme by combining the shearlet shrinkage with complex diffusion strategy. It's proved that the presented smoothing method can successfully remove image noise while preserve both texture and details. To evaluate the noise removing performance of the presented method, three parameters: the peak-to-peak signal-to-noise ratio (PSNR), signal mean squared error (SMSE) and Beta (a parameter used to represent the detail preserving performance) are used. The experiment results acquired from the synthetic and real data prove the good performance of the presented filter.